The hardware is a pair of 15600's with 64GB RAM (MAX), with R80.20 HFA103. Additionally these appliances are running as VSX appliances with 5 active Virtual systems. The following blades are enabled:
fw urlf av appi ips anti_bot content_awareness mon
Tasks: 594 total, 1 running, 593 sleeping, 0 stopped, 0 zombie
Cpu(s): 4.0%us, 1.0%sy, 0.0%ni, 94.4%id, 0.1%wa, 0.0%hi, 0.5%si, 0.0%st
Mem: 65747304k total, 40054152k used, 25693152k free, 1965852k buffers
Swap: 33551672k total, 0k used, 33551672k free, 21309224k cached
Clearly above we see no swap being hit.
08:09:03 up 16 days, 11:37, 1 user, load average: 3.44, 3.29, 3.14
But we also see that the uptime is 16 days, initial memory utilisation was around 16GB after reboot.
total used free shared buffers cached
Mem: 64206 39177 25028 0 1919 20863
-/+ buffers/cache: 16394 47811
Swap: 32765 0 32765
We have at least another 2 virtual systems to implement, assume that each VS will run around 100,000 concurrent connections (200,000 total).
So my concern here is we will not have enough resource to handle this. Additionally I do not see any memory release out side of peak ours when the concurrent connections across all the VS's drop. I would have though that memory should be released.